Prevalence and determinants of dolutegravir resistance among people living with HIV with virologic failure in North-Eastern Uganda: a retrospective cross-sectional study (2022–2024)

Abstract

Background:

In 2016, Uganda adopted the WHO HIV drug resistance (HIVDR) monitoring strategy to track treatment outcomes following the introduction of a dolutegravir (DTG) based regimen as first line among people living with HIV (PLHIV). However, despite widespread transition to DTG-based regimens, evidence on acquired DTG resistance and its determinants remains limited.

Objectives:

To determine the prevalence of acquired DTG resistance among PLHIV with virologic failure and identify associated factors in North-Eastern Uganda.

Design:

We conducted a retrospective cross-sectional study from 24 accredited antiretroviral therapy health facilities in North-Eastern Uganda by 28th February 2025.

Methods:

The study included eligible PLHIV on DTG-based regimens with at least 2 viral loads >1000 copies/mL and a documented HIV genotypic resistance testing result between April 2022 and July 2024. Clinical data were extracted from the national drug resistance dashboard and verified against facility electronic medical records. Logistic regression via generalised estimating equations (GEE) was used to identify factors associated with DTG resistance. A p-value < 0.05 was considered statistically significant.

Results:

Of 225 participants, the median age was 19 years (interquartile range: 15–39), and 55.1% (n = 124) were male. The prevalence of high-level to intermediate DTG resistance was 18.7% (n = 42), with 14.2% (n = 32) having high-level resistance to DTG and 4.4% (n = 10) intermediate resistance. Most participants were unemployed (60.0%, n = 135) and had not disclosed their HIV status (74.2%, n = 167). DTG resistance was associated with presence of nucleoside reverse transcriptase inhibitor (NRTI) drug resistance mutations (DRMs) (adjusted odds ratio (aOR); 12.14, 95% confidence interval (CI): 4.14–35.65, p < 0.001), poor adherence (aOR; 6.19, 95% CI: 2.54–15.11, p = 0.001), DTG use ⩾ 3 years (aOR; 9.19, 95% CI: 1.60–52.74, p = 0.013), and alcohol/substance abuse (aOR; 3.94, 95% CI: 1.74–8.91, p = 0.001).

Conclusion:

Nearly one in five PLHIV with virologic failure exhibited DTG resistance, predominantly high-level. Key associated factors included poor adherence, being on DTG for ⩾3 years, alcohol/substance abuse, and the presence of relevant NRTI DRM. Strengthening family and peer support, integrating substance abuse screening and management in HIV care, and improving adherence may help reduce the risk of DTG resistance. These findings highlight the urgent need for enhanced HIVDR monitoring and tailored interventions to sustain progress toward HIV epidemic control by 2030.

Plain language summary

What causes dolutegravir (DTG) drug resistance in North-Eastern Uganda?

This study aimed to determine the prevalence of acquired DTG resistance among PLHIV with viral load failure and its associated factors in North Eastern Uganda. We analyzed records from 24 health facilities, focusing on people on DTG-based treatment who still had a high viral load and had been tested for drug resistance between 2022 and 2024. Out of 225 people studied, 18.7% had developed resistance to DTG. The main factors linked to this resistance were not taking medicine regularly (poor adherence), being on DTG-based treatment for 3 years or more, and Alcohol or substance use, and also having resistance to the older “backbone” HIV drugs (NRTIs) used with DTG. To keep DTG effective for everyone, HIV programs should strengthen adherence support, screen for and manage substance use, and provide targeted monitoring for those on long-term treatment or with another drug resistance. This is crucial for achieving long-term HIV control.

Introduction

In 2019, the World Health Organisation (WHO) recommended the dolutegravir-based regimen (DBR) as the preferred first-line and second-line regimen for people living with HIV (PLHIV). ¹ This recommendation was driven by DBR’s superior tolerability, lower costs, high potency, and high genetic barrier to resistance, which led to its widespread adoption, particularly in low- and middle-income countries. ² By 2024, more than 25 million people globally were receiving DBR, with Sub-Saharan Africa (SSA) accounting for 85% of all individuals receiving antiretroviral therapy (ART). ³ The massive rollout of DBR increases its population-level exposure, raising concerns about the potential emergence of HIV drug resistance (HIVDR) to dolutegravir (DTG). ⁴

Human immunodeficiency virus (HIV) drug resistance occurs because of changes in the genetic structure of HIV, making it hard for ART to stop it from replicating, and thus increasing HIV related morbidity and mortality. ⁵ The key drivers of drug resistance reported include non-adherence to ART, drug malabsorption, the presence of INSTI-associated resistance mutations such as G118R or R263K, high baseline viral loads, long-term treatment experience, and the use of DTG monotherapy.^4,6 The interaction of these factors has resulted in increasing reports of resistance to this regimen, which is particularly concerning given the wide use of DBR in high HIV burden countries. ⁷ This potential crisis necessitates close monitoring, especially in SSA, where DBR is the most widely used ART regimen. ⁸ A systematic review noted that while global DTG resistance remains low (<4.8%), the situation in SSA is concerning, with one analysis finding that up to 26% of PLHIV had at least a major drug resistance mutation to DTG. ⁹ Another study conducted in PEPFAR-supported low- and middle-income countries, including Uganda, reported a wide range of DTG resistance, from 3.9% to 19%. ⁹ These findings underscore the urgency of understanding and monitoring DTG resistance patterns to prevent a possible crisis in the treatment of PLHIV. ⁹

Uganda, a high HIV burden country with an HIV prevalence of 4.9% and approximately 1.5 million PLHIV, has achieved viral load suppression of 97% for the general population and 89% among children living with HIV by the end of 2025. ⁹ The transition to DBR has been a national priority. ¹⁰ However, a 2018 survey reported an intermediate to high-level DTG HIVDR prevalence of 3.9% among adults (CI: 0.7–7.1). ¹¹ Despite national efforts to monitor HIVDR, region-specific data on the prevalence of DTG resistance and its drivers remain limited. Generating granular evidence from underserved regions is critical to inform tailored interventions and improve treatment outcomes.

The Teso region, with an HIV prevalence of 3.3% as compared to the national average of 4.9%, has one of the lowest viral load suppression rates at 77.6%. ¹² This gap is attributed to nonadherence, drug resistance, and socio-economic challenges such as stigma. ¹³ In line with WHO recommendations, routine drug resistance monitoring is essential to achieve the 95% virologic suppression goal. ⁹ Therefore, this study aimed to determine the prevalence of acquired DTG resistance among PLHIV with virologic failure in the Teso region and identify its associated factors.

Methods

Study design

This retrospective cross-sectional study was conducted using data from twenty-four (24) accredited ART health facilities in North-Eastern Uganda by 28th February 2025. The findings were reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. ¹⁴

Study setting

The study was conducted in Uganda’s Teso region, which has a population of about 2,462,344 according to the 2024 national census. ¹⁵ The region has one of the lowest HIV/AIDS prevalence rates nationally, estimated at 3.3%. ¹² About 52,000 PLHIV are on ART in the region, with nearly 99% receiving DTG-based regimens. The districts in the region are predominantly rural, and most residents are subsistence mixed farmers who keep goats and cattle and tend crops for a livelihood.

Study participants were receiving care from accredited ART clinics in the Teso region. Each ART clinic is led by an ART focal person, typically a nurse or a clinical officer, who is responsible for the day-to-day running of the clinic. The district HIV focal person, who is a trained expert, supervises HIV care and treatment services at the district level.

Study population

This study included all PLHIV on DTG-based regimens who experienced virologic failure between April 2022 and July 2024. PLHIV with virologic failure who were not on a DTG-based regimen were excluded from this study.

Minimum sample size estimation

The minimum sample size of 185 was calculated using Kish Leslie’s formula (1965) for single populations, considering a 95% confidence level, a 5% margin of error, a design effect of 2, a non-response rate of 10%, and a prevalence of DTG resistance of 5.8% based on the HIVDR survey conducted in Tanzania. ¹⁶ To increase the study power, 225 participant records from selected facilities were included in the final analysis.

Sampling approach

All records of PLHIV with virologic failure on a DTG-based regimen were reviewed.

Study variables

The independent variables of the study participants with virological failure considered were those collected at the time of HIVDR testing. They included demographic characteristics (age, sex), employment status, education status (none, primary, secondary), marital status (married, non-married), WHO clinical stage, duration on ART, duration on DTG, nutrition status, adherence to ART, CD4 count, alcohol and substance abuse, psychosocial (PSS) issues (stress, dysfunctional family/client support system, risky sexual behavior, poverty, stigma and discrimination, and disclosure status).

Virologic and treatment history variables included presence or absence of relevant protease inhibitors (PI) drug resistance mutation (DRM), relevant nucleoside reverse transcriptase inhibitors (NRTI) DRM (Abacavir, Tenofovir, Lamivudine, and Zidovudine), and relevant non-nucleoside reverse transcriptase inhibitors (NNRTI) DRM (Efavirenz, Nevirapine, and Etravirine).

The primary outcome

Dolutegravir resistance was defined as the proportion of PLHIV with virologic failure having high-level resistance and intermediate resistance to DTG. The predicted level of DTG resistance was determined by the penalty scores associated with each of the DRMs for DTG present in the results sequence of the patient and categorised based on the Stanford University HIVDR database ¹⁷ :

• Susceptible: 0–9

• Potential low-level resistance: 10–14

• Low-level resistance: 15–29

• Intermediate resistance: 30–59

• High-level resistance ⩾60

Operational definitions

HIV drug resistance

The ability of HIV to reproduce or multiply itself in the presence of antiretroviral drugs. ¹⁸ In Uganda, HIVDR surveillance is conducted routinely in a decentralized, hub-and-spoke model, with samples transported from peripheral facilities to the national health and diagnostic laboratory. Often, two plasma samples are collected from PLHIV with non-suppressed viral load, after at least three (3) consecutive self-reported good adherence scores. When this repeat test still shows a non-suppressed viral load (⩾1000 copies/mL), a drug resistance test is performed. This result is electronically transmitted to the regional hubs for discussion and appropriate action by the HIVDR (third- line) committees.

Acquired HIV Drug Resistance (ADR)

According to the WHO, this develops when HIV mutations emerge due to viral replication in PLHIV receiving ARV drugs. ⁹

ART adherence

This means taking all prescribed antiretroviral medications following the initiation of ART at the correct doses, at the prescribed time intervals and frequencies, and while following any restrictions regarding food and other medications. ⁹

Virologic failure

Two consecutive viral loads ⩾ 1000 copies/mL, done at least 3–6 months apart, with adherence support following the first viral load test.

Inclusion and exclusion criteria

We included records of all PLHIV who were on DTG -based regimens for at least 6 months and had experienced virologic failure, with drug resistance test results. Excluded were PLHIV with virologic failure who were not on a DTG-based regimen, and those without drug resistance results.

Data collection

Using a designed data abstraction tool, clinical and demographic data were extracted from the National HIVDR database and cross-verified against the facility’s electronic medical records to address missing variables.

Nutrition status was assessed using Mid-Upper Arm Circumference (MUAC) and categorized as green (>12.5 cm for children and >21 cm for adults), meaning well-nourished; yellow (11.5 cm to 12.5 cm for children and 18–21 cm for adults) indicating moderate acute malnutrition; red (<11.5 cm for children and <18 cm for adults) indicating severe acute malnutrition.

Adherence to ART in all participants was assessed using self-report and pill count balance and categorised as Good (⩾95% or missed < 2 doses for once daily dosing or ⩽ 2 doses for twice daily dosing), fair (85%–94% or missed 2–4 doses for once daily dosing or 4–8 doses for twice daily dosing) and poor (<85% or missed ⩾5 doses for once daily dosing or ⩾9 doses for twice daily dosing). CD4 count was categorised as <200 or ⩾ 200 cells/mm. ³

Alcohol and substance abuse (ASA) was assessed at the time of collection of the sample for drug resistance using the AUDIT-C screening tool, a 12-point numerical scoring tool. Each parameter is scored on a scale of 0–4. PLHIV with scores of 4–5 were categorised as having ASA, while those with 0–3 scores had no ASA.

Psychosocial (PSS) issues which included stress, dysfunctional family/client support system, risky sexual behavior, poverty, stigma and discrimination, and disclosure status (For the adults disclosure to partner or any family member and for children whether they knew their status and had disclosed to any peer) were identified through a structured assessment by the clinician using HEADSS (Home, Education, Activities, Drugs, Sexuality, Suicide/Depression) framework and clinical care cards questions. ¹⁹ The presence or absence of each specific psychosocial (PSS) issue was directly extracted as a binary variable (yes/no), as recorded in the PSS section of the client’s cards. ¹⁹

The presence or absence of DTG resistance at virologic failure, relevant NRTIs DRM, relevant NNRTIs DRM, and PIs DRM was determined by the centralized laboratory performing the genotype resistance testing and reported in the HIVDR database, where the data were extracted.

Statistical analysis

Data were entered into a Microsoft Excel spreadsheet and exported to Stata Statistical Software: Release 17 (College Station, TX, USA: StataCorp LLC) for cleaning, validation, and analysis. ²⁰ Descriptive statistics of the study participants were summarised using proportions for categorical variables, and medians with interquartile ranges (IQRs) for continuous variables with skewed distributions (age). The association between DTG resistance and the independent categorical variables was determined using the chi-square test, and the association between DTG resistance and the independent numerical variables was determined using the Student’s t-test. To determine factors associated with DTG resistance, a logistic regression using generalised estimating equations (GEE), to account for clustering at the health facility level, was used to estimate adjusted odds ratios (aORs). This approach accounts for clustering by facility through an exchangeable working correlation structure and robust standard errors, providing valid population-level odds ratios while avoiding convergence issues. ²¹ Variables with a p-value < 0.2 in the bivariate analysis and clinically significant based on the existing literature on DTG resistance were included in the multivariate model. ⁹ A p-value < .05 was considered statistically significant. Multicollinearity between the independent variables was absent, as the correlation coefficients were <0.5.

Results

Recruitment of the study participants

Of 273 records retrieved, 13 were excluded because of missing duration information for DTG and ART, and 35 were excluded because of missing drug resistance testing results. A total of 225 PLHIV were included in the final analysis (Figure 1).

Figure 1.

Study flow for prevalence of dolutegravir resistance and associated factors among people living with HIV with virologic failure in North-Eastern Uganda.

Sociodemographic and clinical characteristics

The median age of the participants was 19 years (IQR 15–39), with nearly half (48.9%, n = 110) aged 10–19 years. More than half of the participants (55.1%, n = 124) were males, and the majority (60.0%, n = 135) were not employed. Nearly half (46.2%, n = 104) had received no education. Of all the study participants, 76.0% (n = 171) were non-married, while the majority (88.0%, n = 198) did not have advanced HIV disease stage (Table 1).

Table 1.

Sociodemographic, behavioral characteristics and bivariate analysis for factors associated with DTG resistance among people living with HIV with virologic failure in North-Eastern Uganda, 2022–2024.

Variable	Frequency (n = 225)	DTG resistance		p-Value
		No (n = 183, 81.3%)	Yes (n = 42, 18.7%)
Age categories
0–9	9 (4.0)	6 (66.7)	3 (33.3)	0.271
10–19	110 (48.9)	94 (85.5)	16 (14.5)
20–24	32 (14.2)	25 (78.1)	7(21.9)
25–34	13 (5.8)	12 (92.3)	1(7.7)
>35	61 (27.1)	46 (75.4)	15 (24.6)
Sex
Female	101 (44.9)	85 (84.2)	16 (15.8)	0.326
Male	124 (55.1)	98 (79.0)	26 (21.0)
Employment
Yes	90 (40.0)	76 (84.4)	14 (15.6)	0.328
No	135 (60.0)	107 (79.3)	28 (20.7)
Education status
None	104 (46.2)	84 (80.8)	20 (19.2)	0.974
Primary	100 (44.4)	82 (82.0)	18 (18.0)
Secondary	21 (9.3)	17 (81.0)	4 (19.0)
Marital status
Married	54 (24.0)	43 (79.6)	11 (20.4)	0.712
Non-Married	171 (76.0)	140 (81.9)	31 (18.1)
Advanced HIV disease
No (Stage 1 and 2)	198 (88.0)	163 (82.3)	35 (17.7)	0.302
Yes (Stage 3 and 4)	27 (12.0)	20 (74.1)	7 (25.9)
Nutrition status (MUAC)
Green	204 (90.7)	172 (84.3)	32 (15.7)	<0.001*
Yellow	6 (2.7)	5(83.3)	1 (16.7)
Red	15 (6.7)	6 (40.0)	9 (60.0)
Baseline ART
Nevirapine based	122 (54.2)	100 (82.0)	22 (18.0)	0.976
Efavirenz based	68 (30.2)	55 (80.9)	13 (19.2)
PI based	13 (5.8)	10 (76.9)	3 (23.1)
DTG-based	22 (9.8)	18 (81.8)	4 (18.2)
Duration on ART (years)
<1	12 (5.3)	10 (83.3)	2 (16.7)	0.810
1–4	21 (9.3)	16 (76.2)	5 (23.8)
⩾5 years	192 (85.3)	157 (81.8)	35 (18.2)
Adherence status
Fair	16 (7.1)	15 (93.8)	1 (6.3)	<0.001*
Good	169 (75.1)	149 (88.2)	20 (11.8)
Poor	40 (17.8)	19 (47.5)	21 (52.5)
CD4 count (cells/mm3)
<200	93 (41.3)	77 (82.8)	16 (17.2)	0.499
⩾200	132 (58.7)	106 (79.7)	26 (19.7)
Indication
Failure on the first line	134 (59.6)	110 (82.1)	24 (17.9)	0.735
Failure on the second line	76 (33.8)	60 (78.9)	16 (21.1)
Failure on the third line	15 (6.7)	13 (86.7)	2(13.3)
Stress
No	159 (70.7)	131 (82.4)	28 (17.6)	0.528
Yes	66 (29.3)	52 (78.8)	14 (21.2)
Risk sexual behavior
No	192 (85.3)	155 (80.7)	37 (19.3)	0.575
Yes	33 (14.7)	28 (84.8)	5 (15.2)
Poverty
No	101 (44.9)	82 (81.5)	19 (18.8)	0.960
Yes	124 (55.1)	101 (81.4)	23 (18.6)
Stigma and discrimination
No	154 (68.4)	123 (79.9)	31 (20.1)	0.407
Yes	71 (31.6)	60 (84.5)	11 (15.5)
Alcohol and substance abuse
No	129 (57.3)	115 (89.2)	14 (10.8)	<0.001*
Yes	96 (42.7)	68 (70.8)	28 (29.2)
Dysfunctional family/client support system
No	79 (35.1)	70 (88.6)	9 (11.4)	0.039*
Yes	146 (64.9)	113 (77.4)	33 (22.6)
Disclosure
No	167 (74.2)	133 (79.6)	34 (20.4)	0.269
Yes	58 (25.8)	50 (86.2)	8 (13.8)
Relevant PI DRM
No	167 (74.2)	134 (80.2)	33 (19.8)	0.475
Yes	58 (25.8)	49 (84.5)	9 (15.5)
Relevant NRTI DRM
No	92 (40.9)	89 (96.7)	3 (3.3)	<0.001*
Yes	133 (59.1)	94 (70.7)	39 (29.3)
Relevant NNRTI DRM
No	80 (35.6)	75 (93.7)	5 (6.3)	<0.001*
Yes	145 (64.4)	108 (74.5)	37 (25.5)
Duration on DTG (years)
⩽1	22 (9.8)	20 (90.9)	2 (9.1)	0.067
1–2	114 (50.7)	97 (85.1)	17 (14.9)
⩾3	89 (39.6)	66 (74.2)	23 (25.8)

ART, antiretroviral therapy; DRM, drug resistance mutation; DTG, dolutegravir; NNRTI, Non-Nucleoside Reverse Transcriptase Inhibitors; NRTI, Nucleoside Reverse Transcriptase Inhibitors; PI, protease inhibitor.

^*Significant p-values are given in bold.

Most participants (90.7%, n = 204) had good nutritional status, classified as green on the nutrition assessment scale, and three-quarters (75.1%, n = 169) reported good adherence to ART. More than half (58.7%, n = 132) had a CD4 count ⩾ 200 cells/mm³. At baseline, 54.2% (n = 122) had been initiated on a Nevirapine-based regimen.

Psychosocial assessments showed that 70.7% (n = 159) reported no stress, and over half (68.4%, n = 154) reported no stigma or discrimination. However, 42.7% (n = 96) reported alcohol and drug abuse, and 64.9% (n = 146) came from dysfunctional family settings, and only 25.8% (n = 58) had disclosed their HIV status to their family.

Regarding drug resistance mutations, 74.2% (n = 167) had no protease inhibitor (PI) DRMs, while 59.1% (n = 133) had relevant nucleoside reverse transcriptase inhibitor (NRTI) DRM, and 64.4% (n = 145) had relevant non-nucleoside reverse transcriptase inhibitor (NNRTI) DRM. Most participants (85.3%, n = 192) had been on ART for ⩾5 years, with nearly half (50.7%, n = 114) having been on DTG for 1–2 years, as shown in Table 1.

Bivariate analysis

DTG resistance was significantly higher among PLHIV with poor nutritional status (Red) compared to those with good nutritional status (Green) (60.0% vs 15.7%). Similarly, DTG resistance was higher among PLHIV with poor adherence (52.5%) compared to those with good adherence (11.8%). Those reporting ASA had a higher DTG resistance (29.2%) than non-users (10.8%). Participants from dysfunctional families also had a higher DTG-based regimen resistance (22.6%) than those with stable families (11.4%). In addition, PLHIV with relevant NRTI and NNRTI DRM had higher proportions of DTG resistance, as shown in Table 1.

Prevalence of DTG resistance

Among the 225 participants with virologic failure, the overall prevalence of DTG resistance (defined as high-level resistance and intermediate resistance) was 18.7% (n = 42; 95% CI: 14.1%–24.3%), with the majority (14.2%, n = 32) having high-level resistance to DTG, followed by those with intermediate resistance (4.4%, n = 10), and potential low-level resistance (0.89%, n = 2). The most common DTG mutations are shown in Table 2.

Table 2.

Most prevalent HIV drug-resistance mutations among DTG-resistance cases.

Mutation	Frequency (n = 42)
E138K	34 (80.9%)
G118R	24 (57.1%)
T66A	13 (30.9%)
T66I	11 (26.2%)
L74M	10 (23.8%)
S147G	8 (19.0%)
Q148K	7 (16.7%)
R263K	6 (14.2%)
G140A	7 (16.7%)
G163R	5 (11.9%)
N155H	5 (11.9%)
E157Q	6 (14.2%)
T97A	5 (11.9%)

For each client, there was a sequence of DRM.

DRM, drug resistance mutation; DTG, dolutegravir; INSTI, integrase strand transfer inhibitor; NRTI, nucleoside reverse transcriptase inhibitor.

Multivariate analysis identified several factors associated with DTG resistance. Participants with poor adherence had significantly higher odds of developing DTG resistance than those with good adherence (aOR: 6.19; 95% confidence interval (CI): 2.54–15.11; p = 0.001). PLHIV with alcohol and substance abuse had nearly four-fold higher odds of developing DTG resistance compared to those without such behaviors (aOR: 3.94, 95% CI: 1.74–8.91). Being on a DTG-based regimen for ⩾3 years was associated with a higher odds ratio of having DTG resistance (aOR: 9.19, 95% CI: 1.60–52.74). Finally, the presence of relevant nucleoside reverse transcriptase inhibitor (NRTI) drug resistance mutations was associated with more than 12-fold higher odds of DTG resistance (aOR = 12.14, 95% CI: 4.14–35.65), as shown in Table 3.

Table 3.

Factors associated with dolutegravir resistance among people living with HIV with virologic failure in North-Eastern Uganda, 2022–2024.

Variable	Crude OR	p-Value	Adjusted OR	p-Values
Advanced HIV disease
No (Stage 1 and 2)	Reference
Yes (Stage 3 and 4)	1.91 (0.58–6.27)	0.287	0.21 (0.03–1.31)	0.095
Nutrition status (MUAC)
Green	Reference
Yellow	0.80 (0.04–16.12)	0.886	1.68 (0.38–7.54)	0.496
Red	9.16 (3.15–26.71)	<0.001	3.92 (0.56–27.27)	0.168
Baseline ART
DTG-based	Reference
Nevirapine based	0.93 (0.29–2.91)	0.902	1.33 (0.09–18.56)	0.832
Efavirenz based	1.01 (0.32–3.12)	0.977	0.91 (0.06–13.44)	0.943
PI based	1.33 (0.22–7.84)	0.751	1.45 (0.11–17.77)	0.768
Duration on ART (years)
<1	Reference
1–4	1.68 (0.22–12.76)	0.614	1.34 (0.35–5.18)	0.673
⩾5 years	1.12 (0.31–3.98)	0.85	0.22 (0.03–1.85)	0.164
Adherence status
Good	Reference
Fair	0.62 (0.11–3.64)	0.596	1.91 (0.48–7.51)	0.354
Poor	8.35 (4.48–15.56)	<0.001	6.19 (2.54–15.11)	0.001*
Stigma and discrimination
No	Reference
Yes	0.76 (0.41–1.41)	0.389	1.08 (0.52–2.23)	0.840
Alcohol and substance abuse
No	Reference
Yes	3.45 (1.81–6.57)	0.001	3.94 (1.74–8.91)	0.001*
Dysfunctional Family/client support system
No	Reference
Yes	2.53 (0.91–7.02)	0.073	1.57 (0.34–7.28)	0.562
Disclosure
No	Reference
Yes	0.61 (0.25–1.50)	0.284	1.07 (0.52–2.17)	0.86
Relevant PI DRM
No	Reference
Yes	0.71 (0.40–1.27)	0.256	0.52 (0.23–1.19)	0.123
Relevant NRTI DRM
No	Reference
Yes	12.20 (4.69–31.67)	< 0.001	12.14 (4.14–35.65)	<0.001*
Relevant NNRTI DRM
No	Reference
Yes	5.27 (1.52–18.31)	0.009	3.23 (0.85–12.19)	0.083
Duration on DTG (years)
⩽1	Reference
1–2	1.90 (0.76–4.72)	0.163	2.81 (0.69–11.30)	0.146
⩾3	4.07 (1.37–12.07)	0.011	9.19 (1.60–52.74)	0.013*

ART, antiretroviral therapy; DRM, drug resistance mutation; DTG, dolutegravir; MUAC, Mid-Upper Arm Circumference; NNRTI, Non-Nucleoside Reverse Transcriptase Inhibitors; NRTI, Nucleoside Reverse Transcriptase Inhibitors; PI, protease inhibitor.

^*Significant p-values are given in bold.

Discussion

This study found that poor adherence to ART, alcohol and substance abuse, a duration of DTG-based regimen use ⩾3 years, and presence of relevant NRTI DRM were significantly associated with DTG resistance. These findings suggest that behavioral factors (adherence and substance use), pre-existing drug resistance, and experience of the DTG may coexist with DTG resistance among PLHIV experiencing virologic failure.

The prevalence of DTG resistance among persons with virologic failure in the Teso region was 18.7% (n = 42), which is consistent with the findings from DTG RESIST conducted in seven sub-Saharan countries, ⁹ but higher than national rates (3.9%) reported in a survey of Ugandan PLHIV adults on DTG, ¹¹ and in European and North American cohorts (7.2%). ²² The differences observed may reflect variations in the DTG rollout period, timely access to DTG resistance testing, or the study population characteristics. While this study cannot establish causality, the observed prevalence highlights the importance of ongoing surveillance, as rising DTG resistance among patients having virologic failure could undermine epidemic control efforts.

Poor ART adherence was associated with nearly six-fold higher odds of developing DTG resistance, consistent with studies conducted in South Africa, ²³ Lesotho, ²⁴ and other HIV high-burden countries. ²⁵ A possible explanation for this association is that non-adherence leads to low blood drug levels below effective thresholds, allowing HIV replication and, consequently, the selective survival of viral strains with mutations conferring resistance. ²⁶

Importantly, this study contributes to the growing body of evidence by demonstrating that even in real-world programmatic settings, adherence remains the cornerstone of DTG durability. By quantifying the risk burden in a Ugandan cohort, our findings underscore the urgent need for integrated adherence support interventions and resistance monitoring to safeguard the long-term effectiveness of DTG-based regimens in sub-Saharan Africa.

Alcohol and substance abuse were also associated with increased odds of DTG resistance, consistent with evidence from studies conducted in the United States of America,^27,28 South Africa^29,30 and in Kenya. ³¹ This highlights the role of alcohol and substance abuse in undermining ART adherence, resulting in replication of mutated viruses and consequently increasing the risk of drug resistance.^32,33 Importantly, by demonstrating this association in a Ugandan cohort, our study extends the evidence base beyond high-income settings and underscores the need for integrated adherence and substance-use support services within HIV programs in sub-Saharan Africa. This contribution reinforces the global relevance of behavioral determinants in HIVDR and provides actionable insights for tailoring interventions in high-burden, resource-limited contexts.

A longer duration on a DTG-based regimen (⩾3 years) was associated with higher odds of DTG resistance. This is consistent with findings from Tanzania, ³⁴ Spain, ³⁵ and China.^36,37 Furthermore, a DTG resistance study conducted in sub-Saharan Africa linked the presence of major integrase resistance mutations (e.g., G118R, E138A/K/T) to longer DTG therapy durations, with median durations of 2.4–2.8 years. ⁹ While our cross-sectional design cannot determine temporal sequence, it has been noted that when DTG is initiated, it suppresses the wild-type strain sensitive to ART, allowing pre-existing or newly emerged resistant variants to replicate. ²⁵ These variants can become dominant, under suboptimal drug pressures often caused by non-adherence, drug interactions, and malabsorption, which is associated with sub-therapeutic drug levels due to the diseases among PLHIV with virologic failure.^38,39 However, in this cohort of patients with virologic failure, a longer treatment duration on DTG can also mean prolonged exposure to suboptimal drug pressure, resulting in DTG resistance. By documenting this association in a Ugandan cohort, our study extends regional evidence and underscores the importance of ongoing resistance surveillance and programmatic strategies to sustain DTG effectiveness over time.

The presence of relevant NRTI DRM was associated with DTG resistance; the findings were consistent with those in Brazil, ⁴⁰ Cameroon, ⁴¹ and Liberia, where 41% (13/32) of the PLHIV with virologic failure had intermediate-high resistance to DTG and both NRTIs. ⁴² However, because NRTI and DTG resistance were measured at the same time, the temporal relationship between these resistance patterns cannot be determined. Therefore, our findings should be interpreted as evidence of co-occurrence rather than evidence that NRTI resistance preceded or caused DTG resistance. These results highlight a critical concern that, despite NRTIs forming the backbone of ART in first-line ART regimens such as Tenofovir-Lamivudine (two NRTI) and DTG, their efficacy can be compromised by resistance. ²² By demonstrating this association in a Ugandan cohort, our study adds to the regional and global evidence base, reinforcing the need for routine drug resistance surveillance to guide timely regimen switches. This contribution is vital for sustaining viral suppression and improving patient outcomes in high-burden, resource-limited settings. ⁴³

Limitations and strengths

This study has several limitations. First, it had potential residual confounding; however, this was counteracted with adjustments for clustering and covariates. Second, clients who had not tested for viral load resistance were excluded, limiting the detection rate of resistance mutations, which may have introduced selection bias and may limit the representativeness of the study population. Third, we used a retrospective study design, which limits the ability to establish a causal and temporal relationship. Adherence was assessed using self-report and pill counts, which pose a risk of recall bias and social desirability bias and may have resulted in misclassification of adherence status. We did not conduct baseline DTG resistance testing, making it impossible to determine whether some resistance mutations were present before DTG exposure. The cohort included more children and adolescents, which could affect the generalisability of the findings to broader populations of PLHIV receiving DTG-based therapy. However, the study also has several strengths, including being a multicentre study and using routinely collected data. The use of genotypic resistance testing provides important real-world evidence on DTG resistance among PLHIV with virologic failure in Uganda.

Conclusion

In this study, we found a substantial prevalence of DTG resistance, which was associated with poor ART adherence, alcohol and substance abuse, longer DTG treatment duration (⩾3 years), and the presence of relevant NRTI DRMs. These findings support strengthening adherence support and substance-use screening among PLHIV. They also highlight the importance of routine resistance surveillance among PLHIV receiving DTG-based regimens.

Our study contributes to the literature by quantifying the magnitude of these risks in a Ugandan cohort, thereby extending evidence from other high-burden settings. This reinforces the global relevance of behavioral and biological determinants of DTG resistance and provides actionable insights for programmatic planning in sub-Saharan Africa. We recommend additional studies to determine the ability of a DTG-based regimen in maintaining viral load suppression among PLHIV with pre-NRTI drug resistance mutations in sub-Saharan Africa.